. 5
( 18)


ical damage or personal injury; but strict liability for such damage or injury may
arise from a breach of contract. For example, the liability of a seller to a buyer for
injuries caused by dangerous goods does not depend on negligence, but on breach
of contractual warranty. A seller of goods is generally held to ˜warrant™ that the
goods sold are neither dangerous nor defective. If they cause injury to the pur-
chaser, he or she can sue the seller without having to establish that the seller was
personally negligent. In modern retailing conditions it would often be very di¬cult
to establish that a seller was negligent, because the retailer merely acts as a distribu-
tor of goods that cannot be examined because they are packed in sealed containers,
such as tins or bottles. In other cases the seller could not hope to do more than
ensure the good repute of the maker of goods bought, for they may be products of
which the retailer has no skilled or personal knowledge, such as electronic equip-
ment. Even if the retailer was skilled enough to examine the goods, it would often
be impossible to do so in practice. Thus the liability imposed on the retailer is in
fact a fairly strict one “ much stricter than that on the employer under most indus-
trial legislation. If the goods actually cause damage or injury to the buyer, the seller
can only escape liability in law by showing that the goods were not defective, or that
it was the buyer™s ˜fault™ that the accident occurred. Contributory negligence, as
such, probably does not apply here, for it is a tort doctrine; but if the injured
person™s conduct was su¬ciently foolhardy, a court might reject the claim on the
ground that the buyer ˜caused™ the injuries.
Why should a seller of goods be strictly liable to the buyer for defects when, at
common law (as opposed to statute),24 the manufacturer is only liable to the con-
sumer for negligence?25 At ¬rst sight this is an important question because only
the actual buyer can sue the seller for breach of warranty; if the injury is caused to
a member of the buyer™s family or to a friend to whom the buyer has given the
goods, no such action can be brought, so that many users may have to sue the
manufacturer. However, in practice, the distinction may not be as important as it
seems at ¬rst sight, because a manufacturer will rarely escape liability for negligence

23 See ch. 8.
24 Concerning the position under the Consumer Protection Act 1987, Part I, see 4.8.
25 The manufacturer may be strictly liable in contract, but not to the ultimate consumer who has no
contract with the manufacturer. If a retailer is found liable in contract, the retailer can sue in con-
tract the person who supplied the goods to it, whether an intermediate seller or the manufacturer,
and that contractual liability will be strict in the same way as the retailer™s.
100 Chapter 4

in respect of damage done by a defective product, at any rate if the defect is one in
the manufacture rather than the design of the product.26 Defects are themselves
fairly good evidence of negligence, though they may sometimes be explained away,
and they may sometimes be due to components purchased from other manufac-
turers (for which, at common law, the manufacturer would often not be liable).

4.5 Rylands v. Fletcher, nuisance and animals
Considerations of space forbid any attempt at detailed treatment of these parts of
tort law. It is enough here to note that liability for damage done by dangerous things
escaping from land under the rule in Rylands v. Fletcher may occasionally be ˜strict™,
in the sense that negligence need not be proved. However, the House of Lords has
recently decided that damages for personal injuries cannot be recovered under this
rule;27 and so it is only of marginal relevance to the subject matter of this book. Even
in relation to property damage, it is by no means clear whether, in practice, liabil-
ity under the rule could arise in the absence of negligent conduct. This is not to say,
however, that strict liability for the escape of dangerous things might not be socially
desirable. Witness the Bhopal disaster in India in 1984. But in Australia, the rule in
Rylands v. Fletcher has been ˜subsumed™ into the ordinary law of negligence.
Nuisance is of considerable importance as a tort of ˜strict™ liability, in the sense that
interference with the comfort of neighbours may be a nuisance even if all due care is
taken to prevent that interference. Even here, however, the defendant may in a sense
have been negligent if the concept of negligence is regarded as applying to the whole
activity and not merely to the way it is carried out. When a court says that use of a
particular piece of machinery, for example, is so noisy as to constitute a nuisance even
though every care has been taken to minimize the noise, this is not necessarily an
admission that liability in nuisance is ˜strict™. Such a decision amounts to holding that
the machinery is so noisy that it must not be used at all “ that it is unreasonable to
use it having regard to the harm it caused to the neighbours, notwithstanding the
gain to the person using it. This does not look very di¬erent from the ordinary neg-
ligence formula. The only di¬erence is that it involves application of that formula to
the activity itself, and not to the way it is carried out. As we saw earlier (2.4.9), the
law of negligence, though in theory capable of being applied to negligence in the
whole conduct of an activity “ negligence in design as it were “ is usually con¬ned to
negligence in operation. The peculiarity of the law of nuisance is that the law is con-
cerned just as much with what is done on land as with how it is done.
Be this as it may, nuisance as a source of liability for damage to person or prop-
erty is now almost completely coincidental with negligence. The solitary survival
of strict liability is the rule that an occupier may be liable for physical damage
caused by non-repair of premises even though the occupier neither knew nor had

26 See further 2.4.9.
27 Transco Plc v. Stockport MBC [2004] 2 AC 1.
Departures from the fault principle 101

means of knowing that they were in disrepair.28 Even here, liability is rarely likely
to be strict in practice; non-repair will normally connote negligence.
As to animals, the common law rules of strict liability were so drawn as to
exclude the main forms of damage that animals actually cause, and the Animals Act
1971 has not made much di¬erence in this respect. No doubt the owner of a zoo
from which a lion escapes will be liable for the damage it in¬‚icts; and no doubt also,
the owner of cattle that trespass on a neighbour™s garden or crops will be liable even
if no negligence is proved. But many more accidents, resulting in personal injury,
are caused by dogs or other domestic animals on the roads29 than by escaping lions,
and here there is no liability without, and often not even with, negligence. And
although trespassing cattle may well do a certain amount of damage to crops and
such like every year, cattle do not often trespass on another™s land without negli-
gence on the part of the owner. There are certainly quite a number of accidents
involving bulls, mostly to those employed on farms, but the e¬ect of the Animals
Act 1971 on liability for such animals is obscure.30

4.6 Joint liability
An instance of liability, which in one sense may be said to be ˜strict™, and which is
certainly of more practical importance than the strict liability of animal keepers, is
to be found in the rules relating to ˜joint liability™. This is a technical term, but we
are using it in a non-technical sense to include all cases in which more than one
person is liable for the same damage. There are two main principles. First, all those
responsible for committing a tort in concert are liable for all the damage caused by
the tort, even though it may be possible to identify the contributions of each to the
ultimate damage. Secondly, all those who by their negligence or other fault produce
damage in combination (though not acting in concert), will be liable for all the
damage unless it is possible to identify the separate contribution of each party.
The ¬rst principle applies not only to those actually acting in concert, but also
to cases in which one person assists or encourages another to commit the tort. If
two people agree to beat up a third, then they are both liable in full for the injuries
in¬‚icted, even though it may be possible to identify one assailant as solely respon-
sible for some (or all) of the injuries and the other as responsible for di¬erent (or
no) injuries. This kind of liability is ˜strict™ because the defendant is held liable for
damage which they may not even have caused, in the sense that the damage may
well have occurred even without that person™s assistance. Probably no other form
of tort liability is as ˜strict™ as this.
The second principle covers the situation where two (or more) persons combine
without design to produce damage; where, for example, two motorists by their

28 Wringe v. Cohen [1940] 1 KB 229.
29 See Civil Liability for Animals (Law Com. No. 13, 1967), paras. 36“8.
30 Although the Act abolishes the former principle that an employee could not sue in such cases, the
e¬ect of s. 2(2)(b) and s. 6 seems to be to exclude strict liability for injury done by a bull.
102 Chapter 4

negligence cause an accident in which a pedestrian is injured. It may be impossible
in this sort of case to quantify the damage done by each of the negligent motorists,
and at common law there was no procedure whereby the responsibility could be
shared between the motorists.
In both types of case the common law simply provided that both tortfeasors
were liable in full to the injured person, but the tort victim could choose to sue one
or the other, or both; and likewise, if judgment was given against both, it was left to
the injured person to decide whether to enforce the judgment against one or other,
or both. The law was changed by statute in 1935 (now replaced by the Civil Liability
(Contribution) Act 1978), so that it is now possible for one tortfeasor to claim ˜con-
tribution™ from another tortfeasor liable for the same damage. The amount of con-
tribution is to be assessed by the court according to the share of the responsibility
of the two parties. This change in the law has not a¬ected the tort victim™s position
in any way. The injured party is still entitled to sue both parties, to obtain judgment
in full against both and to enforce judgment against either or both. The only thing
not allowed is recovery by the injured person of more in total than the amount of
damages assessed by the court.
The law of joint liability has become very controversial in recent years. Com-
plaints have come especially from professional groups, such as auditors and archi-
tects, whose job is often to monitor, control or report on potentially dangerous
activities of others. The problem is that where two or more people are jointly liable
for loss, the harmed person will naturally want to sue the party most likely to be
insured against liability for the loss or to have su¬cient resources of their own to
pay any damages awarded.31 If such a party is held liable but cannot recover con-
tribution from any other liable person because the latter lacks the resources to pay
or is uninsured, the party originally sued may end up bearing the whole of the loss
even if their share of the responsibility for it was very small. Various solutions have
been suggested for this type of case, but most involve shifting some or all of the risk,
that one or more of the liable persons will be unable to pay, on to the tort victim.
For this reason, they have been found unacceptable in cases of personal injury.
Some people see the problem of joint liability as a symptom of the failure of the
tort system. From another point of view, however, it simply illustrates that at times
in life, one or another ˜innocent™ party must lose out, and the only question is, who?

4.7 Vicarious liability
The type of liability we have just mentioned is not essentially dissimilar to vicari-
ous liability “ the liability of an employer for the negligence of employees com-
mitted in the course of their employment. This is certainly ˜strict™, in the sense that
the employer is liable, however careful it may have been, although the injured
person must prove loss or damage su¬ered as the result of a tort committed by the

31 Public authorities are a particularly attractive target from this point of view.
Departures from the fault principle 103

employee in the course of the employment. Vicarious liability is discussed in detail
in chapter 9 because it is better understood as a loss distribution device than as a
part of the traditional law of torts.

4.8 Products liability
Part I of the Consumer Protection Act 1987 (which implemented an EC Directive)
is usually said to have introduced a regime of ˜strict™ liability for injury or damage
caused by defective products. Compensation for personal injury32 and damage to
property (other than the defective product) is recoverable under the Act, but
damages for pure economic loss are not. The provisions are complex and we will
not consider them in detail in this book.33 However, a few general comments are in
order. First, the Act does impose strict liability to the extent that under it, not only
the producer of a defective product can be held liable, but also an ˜own-brand™ sup-
plier, an importer and even, in certain circumstances, an ordinary distributor of the
product, regardless of the fact that none of these persons may have been in any way
responsible for the defectiveness of the product. Furthermore, the producer of a
product which incorporates components produced by someone else can be held
liable for defects in component parts even if the producer of the ¬nished product
was in no way at fault, for example, by not doing independent tests on the compo-
nents to ensure their safety.
Secondly, however, there can be liability under the Act only if the product in
question was ˜defective™. A product is defective if it is not as safe as people generally
are entitled to expect. This test is very little di¬erent from the common-law test of
negligence because, in essence, it requires the court to conduct a cost-bene¬t analy-
sis. Indeed, it would seem that the Act has e¬ected little change in the law in this
respect. Take, ¬rst, a case involving a product which is defective because of an iso-
lated malfunction in the manufacturing process. Such a product would undoubt-
edly be defective under the Act regardless of whether the malfunction was anyone™s
fault. A court dealing with such a case under the law of negligence would be very
likely to invoke the doctrine of res ipsa loquitur which, as we have seen, e¬ectively
imposes liability without fault in many cases. So the outcome would probably be
the same under either legal regime. Consider, next, a case involving an argument
that a product was defectively designed. Under the law of negligence, such allega-
tions are very rarely made and the courts are very wary of them. It may be very
di¬cult indeed for an injured person to convince a court that the costs of a partic-
ular product design outweigh its bene¬ts. But things are not likely to be any easier
under the Act. This is very obvious in the case, for example, of drugs. All drugs carry
risks, but this does not mean that all drugs are defective in the terms of the Act: a

32 Strict liability is likely to be of limited value to those who su¬er diseases rather than traumatic
injuries: J. Stapleton, ˜Compensating Victims of Disease™ (1985) 5 Oxford J. Legal Studies 248, 254.
33 See for further details N.J. McBride and R. Bagshaw, Tort Law, 2nd edn (London, 2005), ch. 45.
104 Chapter 4

drug will be defective only if the risks associated with its use make it so unsafe that
they outweigh any bene¬ts it may bring. Nor does it seem that an allegation that a
product was defective by reason of lack, inadequacy or unclarity of instructions or
warnings would be signi¬cantly easier to substantiate in an action under the Act
than in an ordinary negligence action.
Another important respect in which the Act and the common law are essentially
similar concerns the date at which the issue of defectiveness (under the Act) or neg-
ligence (at common law) is to be judged: under the Act the relevant date is the date
the product left the control of the producer, and at common law it is the date of the
alleged act of negligence. This means that liability cannot arise either under the Act
or at common law simply because standards of safety have become higher in the
Thirdly, the Act provides a number of defences, the e¬ect of which is to intro-
duce signi¬cant elements of fault into the liability regime: in particular, contribu-
tory negligence on the part of the user or the injured person; and the so-called
˜state-of-the-art™ or ˜development risk™ defence, which allows a producer to escape
liability by proving that the ˜state of scienti¬c and technical knowledge™ at the time
the product left its control ˜was not such that a producer of products of the same
description as the product in question might be expected to have discovered the
defect™ if it had existed in the product at the time it was under its control.34 This
latter defence is simply a plea of ˜no-negligence™ in the designing, development and
testing of the product.
These features of the new product liability regime put together mean that it is a
regime of strict liability in only a rather weak sense. Two other points should be
made. First, the Act only applies to ˜goods™ (and electricity); it does not apply to ser-
vices, such as the giving of safety advice. Although this limitation on the scope of
the Act is explicable in historical terms, it is very di¬cult to think of any principled
reason why tort liability for defective goods ought to be governed by a di¬erent
regime of rules from that applicable to substandard services. Even more anom-
alously, the Act does not draw a clear distinction between goods and services
because complaints that goods are defective are usually, at bottom, complaints that
someone has performed some service (such as designing the product or supervis-
ing the production line) badly.35
Secondly, by creating a special regime of liability for product-caused injuries, the
Act creates an anomaly between one class of injured persons and other classes (such
as those injured by negligent driving). Why do victims of product-caused injuries36
deserve to be treated di¬erently, and in some respects better than those injured in

34 Consumer Protection Act 1987 s. 4(1)(e).
35 J. Stapleton, ˜Three Problems with the New Product Liability™ in P. Cane and J. Stapleton eds.,
Essays for Patrick Atiyah (Oxford, 1991), 258“70.
36 The Pearson Commission estimated that between 30,000 and 40,000 persons (about 1% of all
victims of accidental injuries) are injured each year in Britain as a result of defects in products
(other than drugs) (Pearson Report, vol. 1, para. 1201). It also estimated that some 5% of these
Departures from the fault principle 105

other ways? Once again, there are historical reasons why the law of product liability
has been singled out for reform “ the Thalidomide tragedy of the 1960s was the major
catalyst of reform of personal injuries law in Britain37 because the litigation which it
generated uncovered serious weaknesses in the law of tort, especially regarding proof
of fault38 and causation. But historical explanations are not justi¬cations. It is not
easy to think of any good reason why victims of product-caused injuries deserve
better treatment than other recipients of tort compensation, let alone the vast major-
ity of injury victims who receive nothing from the tort system.

4.9 Proposals to extend strict liability
4.9.1 Dangerous things and activities
Under the guise of rationalizing and tidying up the present somewhat arbitrary
and haphazard law of strict liability, the Pearson Commission made a set of pro-
posals (in ch. 31 of the Report) which, if implemented, would have very greatly
extended strict liability. The Commission proposed that there should be two new
categories of strict liability introduced. First, the controllers of things or opera-
tions which, ˜by their unusually hazardous nature require close, careful and skilled
supervision™ if the risk of personal injury is to be avoided, should be strictly liable.
Secondly, the controllers of things or activities which, although normally perfectly
safe, are likely ˜if they do go wrong, to cause serious and extensive casualties™,
should also be strictly liable, not merely if there are in fact serious and extensive
casualties, but if there is any injury caused which falls within the risk to be guarded
against. The ¬rst category was designed to cover such things as explosives,
in¬‚ammable gases and liquids; and the second, such things as large public bridges,
dams, major stores and stadiums, and ˜other buildings where large numbers of
people may congregate™.
Unfortunately, the Commission appears to have paid inadequate attention to the
implications of these proposals; nor did it really attempt any serious justi¬cation of
them in principle. It seems clear that the Commission thought they were of rather
minor importance whereas, in practice, their implementation could trigger o¬ a
huge amount of litigation. Even so, some of the limitations inherent in the pro-
posals are striking. Why, for instance, did the second category cover only persons
injured on ˜large™ public bridges, dams and so forth? What conceivable di¬erence
does the size of the bridge, which collapsed on top of her, or underneath her, make
to any rational person? Another di¬culty arises from the proposal to cover build-
ings where large numbers of people may congregate, for this must surely include

may recover some tort compensation (ibid. vol. 1, para. 1202). It appears that the proportion of
product liability claims that actually reach the courts is smaller than the equivalent proportion of
personal injury claims generally.
37 It was the immediate cause of the setting up of the Pearson Royal Commission.
38 It is ironical, indeed, that the 1987 Act includes a development risk defence, because the existence
of such a defence under the ordinary law of negligence cast great doubt on the claims of
Thalidomide victims.
106 Chapter 4

many factories and other workplaces. Yet it is clear that the Commission did not
intend to introduce any new scheme of strict liability for industrial accidents.
The strongest argument against any attempt to rationalize the law along these
lines is the impossibility of providing any workable criterion for deciding what
things or activities are ˜dangerous™. The problem is that by far the greatest number
of accidents are not caused by things normally thought of as ˜dangerous™ at all, but
by everyday things and activities, in particular, motoring. Really ˜dangerous™ things
and activities in fact cause far fewer accidents, no doubt because their use is gener-
ally strictly controlled by regulatory statutes and other precautions. It is also
exceedingly di¬cult to provide any general criterion of dangerousness, because it
is not possible to measure how dangerous a thing or activity is without agreeing
some acceptable accident rate; and it is not easy to compare accident rates for
di¬erent activities and things. How, for instance, can the dangerousness of (say)
bridges be compared with that of motor cars?
The truth is that these proposals of the Commission were ill thought out and
will never be implemented.

4.9.2 Railway accidents
The Pearson Commission proposed a modest extension of strict liability for railway
accidents. Its proposal was that railway undertakings should be strictly liable in tort
for death or personal injury caused wholly or partly by the movement of rolling
stock.39 No justi¬cation was given for these proposals other than the statement that
certain aspects of the operation of railways can be characterized as inherently haz-
ardous. This argument seems to echo those relating to exceptional risks in general,
and it is open to many of the same criticisms. One starts with the paradox that the
movement of rolling stock only causes a minority of the accidents; most railway
accidents arise from activities not normally thought of as hazardous at all, such as
lifting or moving goods and baggage, and slipping or tripping on railway steps.40 It
is impossible to understand why a minority of accident victims should be better
treated in the matter of compensation on the ground that they have been injured
by ˜more hazardous™ activities, when in fact it seems that these ˜more hazardous™
activities actually account for fewer accidents than the ˜less hazardous™ ones. It is
also to be noted that more than half of the injuries occurring on railway lines are
su¬ered by railway sta¬41 who are, of course, entitled to the bene¬t of the industrial
injury side of the social security scheme. Why should these workers also be entitled
to the protection of a strict liability regime merely because railways are hazardous,
when other factory and industrial workers, who also often work with dangerous
machinery, are not? On what possible principle can one justify strict liability for
railway workers but not (say) for coal miners?

39 Pearson Report vol. 1, para. 1186.
40 Ibid., vol. 2, para. 224.
41 Ibid., table 47.
Departures from the fault principle 107

4.10 Ex Gratia compensation schemes
Confronted with claims for compensation based on allegations of fault such as
could form the basis of a tort claim, governments sometimes react by establishing
schemes the aim of which is to bypass the tort system in favour of an administrative
process for assessing and paying compensation. In one sense, such schemes can be
seen as an application of the fault principle. But typically, the motivation for such
schemes is not an acceptance of responsibility for fault but rather a desire (for
reasons such as political expediency, or ˜sympathy™ for the victims whose plight is
particularly heart-rending) to provide compensation regardless of fault. In this
sense, such schemes can be understood as involving the acceptance of a form of
strict liability. The schemes considered here can be distinguished from special
schemes established as adjuncts to the social security system for the bene¬t of
speci¬c classes of people (13.6, nn 104 and 105). The basic aim of such special
schemes, as of the social security system generally, is to meet need. By contrast, the
schemes considered in this section are usually understood in terms of discharging
some sort of responsibility owed by government to the injured.

4.10.1 Vaccine damage
A very small proportion of children su¬er severe brain damage as a result of vacci-
nation, in particular, vaccination against whooping cough. The Association of
Parents of Vaccine Damaged Children told the Pearson Commission that it had reg-
istered 356 cases of vaccine-damaged children. The Association pleaded for some
form of strict tort liability to be introduced to help such children, mainly on the
ground that child vaccination has for many years been recommended by the gov-
ernment. Moreover, vaccination is a classic case of the ˜free-rider™ problem much
discussed by economists. The bene¬t to each individual child of being vaccinated
will not be very great provided most other children are vaccinated, thus greatly
reducing the risk of infection; yet if the parents of all children reasoned in this way,
vaccination would decline and the diseases in question would spread more widely
again, with greater risk to all. Unlike the administration of many drugs, vaccination
is designed not only to bene¬t the recipient “ although it may do this “ but also to
bene¬t other vulnerable members of the population. To this extent, we may think
that young children who are vaccinated before they are old enough to understand
the issues are being used for the bene¬t of others.
The imposition of strict liability on drug manufacturers does not meet the
problem because it is typically not clear that the vaccine itself is defective. The
Pearson Commission therefore proposed the imposition of ˜strict liability™ in tort on
the government or a local authority where a vaccine was given following a recom-
mended programme for which the government or local authority was responsible.
However, the political pressure on this issue was so great that the government felt
forced to announce some concession even before the Commission reported, and it
promised a lump sum payment of £10,000 to any child who could be shown to have
108 Chapter 4

been ˜severely disabled™ as a result of a vaccination against various ailments.42 E¬ect
was given to this promise in the Vaccine Damage Payments Act 1979, though this
was treated as an interim measure without prejudice to the possible acceptance of
full government liability in tort. Such acknowledgement of liability was never made,
and in 1985 the Secretary of State was empowered by legislation to increase the
amount of the lump sum payment.43 It was increased to £30,000 in 1988 and to
£100,000 in 2000.44
Despite the arguments in favour of compensation for vaccine damage outlined at
the beginning of this section, it is not obvious that a small number of children dis-
abled in this particular way should be singled out for especially generous treatment.
The OPCS Disability Survey estimated that there were some 136,000 children under
16 in the four most serious disability categories; and the Pearson Commission esti-
mated that 90% of severely disabled children were disabled from birth. Furthermore,
it is very di¬cult to establish a causal connection between vaccination and disabil-
ity because small children not infrequently develop convulsions for the ¬rst time in
the ¬rst 2 years of life, and only some of these attacks follow routine vaccinations.45
In fact, however, after an initial wave of applications immediately after the estab-
lishment of the scheme, very few awards have been made since 1988, and most appli-
cations fail on the causation issue. Moreover, although the scheme is generous to this
group of disabled children, £100,000 is only a fraction of what a seriously disabled
child could expect to recover as the result of a successful tort claim.46

4.10.2 HIV
In 1988 the government established the Macfarlane Trust to administer compensa-
tion payments to haemophiliacs infected with the HIV virus as a result of receiving
contaminated blood products. The initial amount made available was £10 million;
but as a result of continued lobbying by the Haemophilia Society and on the basis
of an undertaking to exclude the government from liability, another £14 million was
added to the fund in 1990. A further £44 million was added in 1991, and £15 million
more between 1993 and 2001. Because the government has never accepted liability
for HIV infection resulting from contamination of blood products, and no liability
has been established by judgment of a court, this scheme operates e¬ectively as a
no-fault compensation scheme.

4.10.3 Hepatitis C
In the 1970s and 1980s several thousand people were chronically infected with
Hepatitis C, once again as a result of being treated by the NHS with contaminated

42 ˜Severely disabled™ is de¬ned in terms of departure from ˜normality™ for a child of the relevant age “
originally 80% departure, and now 60%.
43 Social Security Act 1985, s. 23.
44 Vaccine Damage Payments Act 1979 Statutory Sum Order 2000.
45 Pearson Report, vol. 1, para. 1389; R. Goldberg,˜Vaccine Damage and Causation “ Social and Legal
Implications™ (1996) 3 JSSL 100.
46 S. Pywell, ˜The Vaccine Damage Payment Scheme: A Radical Proposal™ (2002) 9 JSSL 73.
Departures from the fault principle 109

blood products. In 2001 a court held the government liable to pay compensation,
ranging from £10,000 to more than £200,000, to a sub-group (but not all) of these
people. For some time, the government resisted calls to compensate others, but in
2003 set up the Skipton Fund ˜without admission of legal liability™ because, it was
said, ˜on compassionate grounds this was the right thing to do™. Quali¬ed applicants
receive a basic payment of £20,000, and applicants who have contracted cirrhosis or
liver cancer are entitled to a further payment of £25,000. Applicants who were
treated with certain speci¬ed products do not have to prove a causal link between
their illness and the treatment. Applicants treated with other possibly contaminated
products are dealt with ˜on a case-by-case basis™. Payments received from other
sources (such as tort compensation) do not reduce the sum awarded. The notewor-
thy thing about this scheme is that it envisages the payment of compensation to
people who might have been infected in circumstances that were held not to attract
tort liability. Once again, political pressure bene¬ts a group who undoubtedly
deserve sympathy but whose plight is no di¬erent from that of many other disabled
people. Whereas compensation payable to those who were successful in court was,
of course, related to the severity of the e¬ects of the contamination on each indi-
vidual, the Skipton Fund payments only discriminate between victims on the basis
of whether or not they have a particular illness.

4.10.4 Variant CJD
In 2001 the government established a trust to compensate victims of variant CJD,
the human form of BSE. The Trust Deed states that the government ˜wishes to
provide funds in such a manner as does not prohibit [su¬erers] or their families
from taking legal proceedings against the Crown and/or related bodies if so advised
but wishes to ensure so far as possible that in the event of such proceedings being
brought the sums paid [under the Trust Deed] are taken into account in the com-
putation of damages to be claimed in any such proceedings™. Payments under the
scheme are modelled on common law damages; but guidance issued by the vCJD
Trust states that the compensation ˜is in many respects greater than the payment of
damages which would be awarded by a Court™. The assumption seems to be that the
government may be liable; and by providing generous compensation, it is obviously
hoped that litigation will be averted. The government has pledged more than £67
million to the fund to compensate an anticipated 250 claimants.

Causation and remoteness of damage

5.1 Introduction
A person cannot incur tort liability to pay damages for injury or damage su¬ered
by another unless that injury or damage was caused1 by the former™s tortious
conduct. This is as true of strict tort liability as it is of fault-based tort liability.
Causation of harm is essential to tort liability because tort law is a set of princi-
ples of personal responsibility for conduct. Tort law compensates the injured,
but only if someone else was responsible for those injuries; and normally a per-
son will not be responsible for injuries unless their conduct caused the harm. In
other words, the tort system is a ˜cause-based™ compensation system.2 These decep-
tively straightforward statements raise complex issues which are usually dealt
with by considering two questions: ¬rst, did the tortious conduct in fact cause
the damage? Secondly, whatever the answer to the ¬rst question, ought the tort-
feasor to be held liable for the loss su¬ered by the injured person? If the answer
to the ¬rst question is ˜no™, then the answer to the second will usually, but not
invariably, also be negative. But answering the ¬rst question a¬rmatively by
no means always leads to the imposition of liability. The reason for this is expres-
sed by the courts in a variety of ways: sometimes by saying that the damage
was not foreseeable; sometimes by saying it was too ˜remote™; sometimes by
saying the damage su¬ered is not of a kind recognized by the law; sometimes by
saying that the defendant™s negligence was not the ˜real™ or ˜proximate™ cause of
the damage. We will consider these di¬erent formulations in turn later in the
chapter; but, ¬rst, we must examine what it means to say that a tort was a ˜factual
cause™ of harm.

1 The simplest type of causation is illustrated by a case where a negligent driver hits a pedestrian.
Tort law also recognizes other types of causation as being su¬cient for liability. Where tort law
imposes liability for nonfeasance, the causal connection between the omission and the harm is
failure to prevent the harm occurring. An employer may be held vicariously liable for injury
resulting from an employee™s tort committed in the course of employment even though the
employer did not cause the injury but only created the opportunity for the employee to cause it.
Yet other forms of causal connection recognized in tort law include inducing or assisting another
to cause injury. For the sake of simplicity, the discussion in this chapter for the most part con-
centrates on causing in the ¬rst simple sense.
2 The implications of this statement are explored in more detail in ch. 19.

Causation and remoteness of damage 111

5.2 Factual causation
5.2.1 Proving causation
Generally speaking a person cannot be held liable in tort unless it can be said that
˜but for™ that person™s tort, the harm complained of would not have occurred; or, in
other words, that the tortious conduct was a necessary condition of the harm; or,
di¬erently again, that the tortious conduct caused or contributed to the harm. We
must add the words ˜or contributed to™ because, of course, a number of actions may
combine to cause damage as, for example, in many road accidents where both
drivers are at fault.
In the great majority of cases, this requirement of ˜but-for™ or ˜factual™ causa-
tion gives rise to no practical di¬culties. Indeed, in the typical case of perso-
nal injury it is easier to determine whether conduct was a factual cause of harm
than whether that conduct was negligent. In practice, an injured person who
is looking for someone to sue will normally ¬nd that person by looking for a fac-
tual cause of the harm rather than for negligent conduct. A person who is run over
in the road will normally start by blaming the driver of the car that did the dam-
age; the driver was obviously a cause of the accident. Whether the driver was also
at fault is a di¬erent and often more di¬cult question. But there are several types
of case in which the issue of factual causation may present di¬culties.
In the ¬rst place, accidents sometimes occur without the cause being immediately
apparent. A soft-drink bottle explodes in someone™s face, or a person dies while being
operated upon, or a baby is born with brain damage; an aircraft falls out of the sky,
or a ferry sinks in calm water with the loss of many lives; a ¬re starts in a crowded
football stadium and many are killed or injured in the rush to escape. Much investi-
gation may be required to ascertain the causes of such incidents. In other cases we
may know that there has been negligence, but be uncertain at ¬rst whether the person
guilty of negligence contributed to the damage. A worker, for example, is found dead
near some machinery, which has been negligently maintained in a dangerous manner
by the employer. The employer is guilty of negligence, but has that negligence caused
or contributed to the worker™s death?3 Similarly, we may know that an injured person
used a defective product (or took a defective drug) for which its producer is ˜strictly
liable™, but not be con¬dent that the defect caused the injuries. In these types of case,
too, a detailed investigation may be required to discover the truth. Such an investi-
gation would obviously be a formidable burden for an injured person to have to
arrange, let alone ¬nance. In the case of a large-scale public disaster involving
many deaths or injuries, the injured will usually be relieved of such a burden because
public inquiries are invariably held under statutory powers to investigate the causes
of such accidents. In cases of workplace accidents, the health and safety inspectorate
may help to investigate the causes of the accident; but in cases of medical mishap, for
example, such publicly funded assistance is unlikely to be available.

3 See e.g. Caswell v. Powell Du¬ryn Collieries [1940] AC 152.
112 Chapter 5

Sometimes the injured person may be helped by the doctrine of res ipsa loquitur
which, as we have seen, says that where an accident results from a situation or thing
under the exclusive control of X or persons for whom X is responsible; and where,
further, the accident is such that, in the ordinary course of events, it would not
occur without negligence on X™s part, it is for X to explain how it happened and that
it was not caused by their tortious conduct.

5.2.2 Causing and increasing the risk of harm
Proving factual causation may be very di¬cult in many cases involving disabilities,
diseases or illnesses as opposed to accidents causing traumatic injuries.4 Sometimes
these problems arise from the fact that medical knowledge about the causation of
many conditions is quite limited and because many diseases are the result of a
complex combination of factors that interact in unknown ways. In other cases, it may
be possible to say that conduct of the allegedly tortious kind is a cause of a particu-
lar condition in a certain percentage of cases, but it may not be possible to say
whether it caused the condition in this case. The ¬rst source of di¬culty becomes less
acute as medical science advances. For example, much more is known today about
the causes of heart disease or cancer than was known 40 years ago. But the second
source of di¬culty may be impossible to eradicate because many conditions can be
caused by more than one factor.
Such di¬culties of proof are to some extent ameliorated by the fact that the law
only requires proof that the injury or damage was ˜more probably than not™ caused
by the negligence: in other words, that more probably than not, the harm would not
have occurred but for the tort. In one case5 the House of Lords appeared to go
further by holding that a claimant could recover damages in respect of a skin con-
dition even though it could not be proved on the balance of probabilities that the
defendant™s negligence caused the condition, but only that it ˜increased the risk™
(albeit substantially) that the claimant would contract the condition. A major
problem with this approach is that a defendant responsible for a harmful process
(e.g. one involving the use of some carcinogen) or product may have to pay damages
to a large number of people, many of whom contracted their disease from some
other source than the defendant™s tortious conduct.6 This may partly explain why
the House of Lords later reinterpreted its earlier decision and said that it was in fact

4 J. Stapleton, ˜Compensating Victims of Disease™ (1985) 5 Oxford J. Legal Studies 248, 250“2, 267 n.
54; Disease and the Compensation Debate (Oxford, 1986), ch. 3.
5 McGhee v. National Coal Board [1972] 3 All ER 1008.
6 It must be said that this objection also applies, although less strongly, to the ˜balance of pro-
babilities™ test of causation. Under this test it is theoretically possible for a claimant to recover
damages from a defendant whose negligence did not cause the claimant any loss. The justi¬-
cation for the test is that reconstructing the past is a very di¬cult exercise, and that a high
level of certainty is often unattainable. In the criminal law, however, defendants can be con-
victed only if they are guilty ˜beyond reasonable doubt™. As we know only too well, how-
ever, it is not unknown for the innocent to be convicted as a result of purported application of
this test.
Causation and remoteness of damage 113

a case in which the court was satis¬ed that the tortious conduct had more probably
than not caused the skin condition.7
However, more recently the House of Lords has returned to its earlier approach
in a case of major signi¬cance involving a disease called mesothelioma.8 It is known
that mesothelioma is caused (only) by exposure to asbestos, and that the risk of con-
tracting the condition increases with increased exposure. Each of the claimants in
this case had been tortiously exposed to asbestos in more than one workplace, and
it was possible to say that each exposure had increased the risk of mesothelioma. But
it was not possible to say, in relation to any of the claimants, which exposure(s) had
triggered their illness or even by how much any particular exposure had increased
the risk of their contracting the illness. The House of Lords held that all the employ-
ers were liable for the harm resulting from the illness because each had ˜materially™
increased the risk of mesothelioma. This decision clearly establishes that in some
cases, it will not be necessary for a tort claimant to prove, more probably than not,
that the harm they su¬ered was caused by the tort. All that need be proved is that
the tort ˜materially™ increased the risk of the harm. Unfortunately, the House of
Lords gave very little guidance about the sorts of case in which this exception to the
normal balance-of-probabilities rule of causation will be applied. Unless the bound-
aries of its application are clearly speci¬ed, there is a danger that it will eventually
swallow up the basic rule. At the very least, uncertainty about the boundaries is likely
to generate litigation.
Another possible line of approach to such di¬culties (in some cases, anyway) is
to argue that the tortious conduct increased the risk, that the victim would contract
a particular disease or su¬er a particular condition, by a particular amount (say,
25%) and to claim damages proportional to the increase of risk. (In cases where the
material-increase-in-risk rule applies, the victim is compensated in full for all the
harm attributable to the tortious risk-increasing conduct.) This would mean that if
the victim su¬ered losses valued at, say, £100,000, the claim would be for £25,000.
Such an approach might be attractive where, for example, a particular medical pro-
cedure has a 25% success rate, but the defendant doctor negligently fails to carry out
the procedure in the victim™s case, thus depriving the victim of a 25% chance of a pos-
itive health outcome. When it ¬rst considered this type of claim, the House of Lords
rejected it on the basis that damages are payable only if the tortious conduct
increased the risk of harm by more than 50%; or, in other words, if the tortious
conduct more probably than not caused the harm.9 If it did, the victim would be enti-
tled to damages for all the harm attributable to the tort. But if all the victim could
prove is that more probably than not, the tortious conduct increased the risk of harm
by less than 50%, nothing could be recovered. This means, in e¬ect, that damages for
(and proportional to) loss of a chance cannot be recovered in a personal injury tort

7 Wilsher v. Essex AHA [1988] AC 1074. See generally J. Stapleton, ˜The Gist of Negligence, Part II™
(1988) 104 LQR 389.
8 Fairchild v. Glenhaven Funeral Services Ltd [2003] 1 AC 32.
9 Hotson v. East Berkshire HA [1987] AC 750.
114 Chapter 5

claim. The House of Lords has since rea¬rmed this approach in a case where delay
in diagnosing cancer reduced the claimant™s chance, of surviving cancer-free for 10
years, from 42% to 25%.10
On the face of things, it is di¬cult to see how the reasoning in these loss-of-chance
cases can be reconciled with the material-increase-in-risk approach. If a person, who
can prove (on the balance of probabilities) only that tortious conduct ˜materially™
increased the risk of harm, can recover damages for that harm, why should a person,
who can prove (on the balance of probabilities) that tortious conduct deprived them
of a speci¬ed (signi¬cant) percentage chance of avoiding harm, not recover damages
proportional to the increase in risk? As things stand at the moment, it seems that a
person who can quantify the increase in risk and asks for damages proportional to
that risk is worse o¬ than a person who cannot quantify the increase in risk and
asks for damages for the harm. We could also turn the question the other way round:
if all the claimant can prove is that the tortious conduct increased the risk of harm
by less than 50%, why should damages for the harm “ as opposed to damages pro-
portional to the increase in risk “ be recoverable? The basic balance-of-probabilities
rule strikes a balance between the interests of doers and su¬erers of harm: provided
the victim can prove that the tortious conduct was more probably than not a fac-
tual cause of the harm, damages can be recovered for that harm. But we might
think that if the balance-of-probabilities requirement is relaxed (to the victim™s
bene¬t) in favour of the material-increase-in-risk requirement, tortfeasors should
be given some corresponding concession, such as proportionate liability.11
However acceptable proportionate liability may seem in cases where a single
tortfeasor materially increases the risk of harm, we might be less happy about it in
cases involving more than one tortfeasor (such as the mesothelioma case). Suppose
a drug with adverse side-e¬ects is manufactured and marketed by several com-
panies and taken by a large number of people. If victims sue the manufacturers in
tort, individuals may not be able to prove which of the several manufacturers made
and marketed the particular pills (or whatever) taken by them. One solution would
be to impose liability on each manufacturer according to its market share.12 This
would be roughly equivalent to imposing liability proportional to increase in risk.
The problem, from the point of view of victims, is that unless they can successfully
sue and recover from all the manufacturers, they will not be fully compensated. The
rationale of the basic rule of joint liability (see 4.6) is precisely to protect victims
from such an eventuality. In some jurisdictions, proportionate liability schemes
have been introduced, but typically they do not apply to personal injury cases.
The complexity of the issues of causation that arise in the sorts of cases we have
been considering may suggest that we have here reached the limits of the practical
utility of tort law as a system of compensating victims of personal injury. However,

10 Gregg v. Scott [2005] 2 WLR 268.
11 At the time of writing, a case raising this issue is before the House of Lords: Barker v. St Gobain
Pipelines Plc [2004] EWCA Civ 545.
12 See J.G. Fleming, The American Tort Process (Oxford, 1988), 258“60.
Causation and remoteness of damage 115

it must be said that such problems of causation are not unique to the tort system “
they can arise in any system under which entitlement to compensation depends on
tracing a causal connection between some event and the loss su¬ered. Nor do the
problems end here.

5.2.3 Omissions
Suppose that an employer negligently omits to provide safety belts for workers
who are working at such dangerous heights that belts should be worn. A worker
falls and is killed; if they had been wearing a safety belt the fall would not have
occurred. Is the employer™s omission to supply the safety belt a factual cause of the
worker™s death? In order to answer this question we must ask whether, but for the
omission, the worker would have died; but we cannot answer that question unless
we know (amongst other things) whether the worker would in fact have worn the
belt if it had been provided. Because of the hypothetical nature of this latter ques-
tion (it is not about what happened but about what might have happened), it may
in many cases be extremely di¬cult to answer with con¬dence. As a result, the
outcome of a case that raises such a question may in the end depend on whether
the court requires the victim to prove that the precaution would have been taken
or the defendant to prove that it would not. In McWilliams v. Sir William Arrol &
Co.13 the House of Lords held that the normal rule that the claimant bears the
burden of proof applies in such a case. In policy terms this approach is arguably
unsatisfactory because it gives employers inadequate incentives to perform their
duty to provide safety precautions and to see that they are used. This seems to have
been appreciated in Bux v. Slough Metals Ltd 14 where an employer was held liable
for the employee™s failure to wear safety goggles on the ground that he probably
would have worn them if there had been adequate instruction and supervision.
In practice, courts may be prepared (as was Lord Reid in McWilliams) to presume,
in the absence of some evidence to the contrary, that the victim would have used
any safety device which it was generally considered reasonable to use. This appro-
ach e¬ectively turns the question: would the victim have used the safety device if it
had been provided? into the question: would the reasonable person in the victim™s
position have used the safety device if it had been provided?15 The former, we might
say, poses the causal question in a ˜subjective™ way, and the latter in an ˜objective™ way.
The choice between the subjective and the objective approaches is also important in
cases where the allegedly tortious conduct is failure to warn: for instance, where
doctors fail to notify patients of risks inherent in medical procedures,16 or product
manufacturers fail to warn users of risks inherent in the product.17 In practice,

13 [1962] 1 All ER 623.
14 [1974] 1 All ER 262.
15 This question is really equivalent to asking: should the victim have used the safety device?
16 E.g. Chester v. Afshar [2005] 1 AC 134.
17 E.g. Hollis v. Dow Corning Corp. (1995) 129 DLR (4th) 609 (subjective approach adopted). In
Smith v. Arndt (1997) 148 DLR (4th) 48 the question was whether a woman would have had an
abortion if she had known of a risk of injury to her foetus (objective approach adopted).
116 Chapter 5

however, the choice may not be so important as it appears at ¬rst sight. This is
because evidence given by the victim personally about what they would have done if
the tortious conduct had not occurred is likely to be tainted by self-interest and after-
the-event rationalization. For this reason, courts are likely to give it little weight and
to decide the causation issue mainly on circumstantial evidence, even if the question
to be answered is framed subjectively. This approach e¬ectively objectivises the cau-
sation test.

5.2.4 Multiple causal factors
The but-for test causes di¬culty in cases involving multiple causal factors. Two
types of case can be usefully distinguished. A classic example of the ¬rst type is that
of two ¬res, independently started by A and B respectively, which unite and spread
to C™s house which is destroyed.18 If we ask whether A or B in fact caused the damage,
the but-for test would seem to acquit both A and B of liability.19 The damage would
have occurred even without A™s conduct and also without B™s conduct; but it is gen-
erally conceded to be unfair to let both parties escape liability. A similar conundrum
is raised by consecutive causal factors, either of which would be su¬cient on its own
to bring about the result in question. A runs over C, wounding him in the leg; later
B shoots at C, in¬‚icting a wound which necessitates immediate amputation of the
leg and would have done so even if there had been no earlier wound.20 Here again
we cannot say that but for A™s negligence, or B™s shot, C would have had an unin-
jured leg. But again it would be unfair to acquit both A and B of having caused the
injury. In this type of case the outcome often depends on whether or not the causal
factors operated more or less contemporaneously: if so, both will be held responsi-
ble; if not, the ¬rst in time will be held liable21 on the basis that a person cannot be
said to have ˜caused™ an injury which has already been su¬ered.
A second and slightly di¬erent type of case occurs where one of the causal factors
is hypothetical. Suppose, for instance, that C is killed in a car accident caused by B™s
negligence while being driven to an airport to catch a plane. If it had not been for B™s
conduct C would not have been killed in the car accident: but suppose that the plane
which C was to catch subsequently crashes with the loss of all on board. Would we still
say that C would not have been killed but for B™s conduct? In a US case,22 a boy fell
from a bridge to what was certain death or grave injury below, but in his fall he came
into contact with some high tension wires negligently maintained by the defendant,
and was electrocuted. Was the defendant™s conduct a factual cause of the boy™s death?
As in cases of the ¬rst type, where there is any signi¬cant interval of time between the

18 See Kingston v. Chicago & NW Railway (1927) 22 NW 913.
19 A way of overcoming this problem is to adopt the so-called ˜NESS™ test of factual causation which
classi¬es something as a cause if it was a necessary element of a set of conditions which together
were su¬cient to cause the outcome in question. In the example in the text, application of this
test would result in both A and B counting as factual causes.
20 Baker v. Willoughby [1970] AC 476.
21 Performance Cars v. Abraham [1961] 3 All ER 413 (property damage).
22 Dillon v. Twin State Gas & Electric Co. (1932) 163 A 111.
Causation and remoteness of damage 117

occurrence of the damage as things actually turned out, and its probable occurrence
as things might have turned out, here too the courts are apt to treat causal connection
as established. In Burmah Oil Co. v. Lord Advocate 23 for instance, where British armed
forces had destroyed oil installations in Burma to prevent their falling into Japanese
hands, it was not questioned that the acts of the armed forces were the ˜cause™ of the
¬nancial losses in¬‚icted on the owner of the installations.
Any feeling of unfairness to the defendant in such cases may be dealt with in
assessing damages. Thus, in the US case it was held that the defendant did cause the
boy™s death, but that the damages awarded must be calculated on the footing that
the boy probably had only a few seconds to live, or at best would have been gravely
injured. Similarly, in the Burmah Oil case, it is probable that any compensation
awarded would have taken account of the fact that the installations were about to
fall into the hands of an invading enemy army when they were destroyed.24 In assess-
ing damages for future economic loss the courts regularly speculate about the likely
occurrence of events that would increase or reduce that loss, and award damages
proportional to the likelihood. In one case, for example, a worker was burnt on the
lip as a result of his employer™s negligence; the burn triggered o¬ a pre-malignant
condition which the worker had before he was burnt, and he contracted cancer. The
employer was held liable for the cancer as well as for the burn, but the damages were
reduced to take account of the fact that the worker might have contracted cancer
even if he had not been burnt.25
This assessment-of-damages approach seems a fair way of dealing with relevant
hypothetical events. But courts seem willing to apply it only to hypothetical future
events, not to hypothetical past events.26 For instance, in a case like McWilliams v.
Arroll (5.2.3), why should the victim not receive damages calculated according to
the chance that they would have used the safety belt? Why does the law usually
require proof on the balance of probabilities, that the tortious conduct caused the
harm complained of, before it will award damages re¬‚ecting the chance that the
victim would have su¬ered harm even if the tort had not occurred? One answer
may be that the courts wish to discourage litigation in cases where the tortfeasor™s
contribution to the harm is very hard to assess with any con¬dence. But even in
cases where it is possible to quantify that contribution (as in Hotson or Gregg v. Scott
(5.2.2)), there is great resistance to imposing liability proportionate to risk.
In cases (of the type discussed in the ¬rst paragraph of this section) involving
consecutive causal factors, an approach similar to the assessment-of-damages tech-
nique is adopted by applying the so-called ˜vicissitudes™ principle, which says that
where events relevant to the assessment of damages occur before the trial, the court
will take those into account. For example, suppose a worker sustains a back injury

23 [1965] AC 75.
24 [1965] AC 75, 112“13, 165.
25 Smith v. Leech Brain [1962] 2 QB 405.
26 Unless the hypothetical event in question is some act or omission of someone other than the
claimant and the defendant: Allied Maples Group Ltd v. Simmons & Simmons [1995] 1 WLR 1602.
118 Chapter 5

at work through the negligence of the employer and later contracts a condition that
would in any event have caused incapacitation at least as great as the injury. In such
a case the employer would not be required to pay damages for the period after the
onset of the condition.27

5.3 Limits on the liability of factual causes
Even where it is clear that the tortfeasor™s conduct was a factual cause of the victim™s
harm, the tortfeasor may be held not liable for damage that has occurred in an
unexpected or unusual or unforeseeable way. This limitation on liability covers two
somewhat di¬erent situations. First, it prevents a person being held liable for the
consequences of a tort where these are ˜too remote™ in time and space from the tor-
tious conduct, and in particular, where some other event intervenes between the
tortious conduct and the occurrence of the harm; for example, where A injures B
in a road accident and B is injured again in a further accident while being taken to
hospital. Once a person has been injured by the negligence of another, they may in
time to come su¬er further injuries or accidents which might not have occurred at
all ˜but for™ the original negligence; but it is generally felt that it would be unfair to
hold the tortfeasor liable for all such consequences. Secondly, this limitation on lia-
bility saves a person from liability even for damage which follows tortious conduct
closely in time and space, but which occurs in an unusual or freakish way or is of
an unexpected kind; for example, where a plank dropped into a ship™s hold starts a
con¬‚agration which destroys the whole ship.
The policy underlying the denial of liability in cases of this kind is clear enough
when the defendant™s liability is based on negligence: it is not reasonable to expect a
person to take precautions against freakish or unexpected or unusual events. But
even in cases where liability for injury or harm is strict, we may think it unfair to
hold a person liable for such events: people ought to be in a position to take account
of their potential legal liabilities in advance in deciding what activities to engage in
and on what scale; and a person cannot reasonably be expected to take account of
freakish or unexpected or unusual events. The technical or ˜conceptual™ shape of this
part of the law is a morass. At least three reasons have been given for denying liabil-
ity for freakish or unexpected outcomes. These are, ¬rst, that the tortious conduct
was not the ˜cause™ of the damage; secondly, that the damage was not within the risk
required to be guarded against; and, thirdly, that the damage was not foreseeable.

5.3.1 Legal causation
The but-for test is very indiscriminate in that it will identify as causes many factors
that are of little interest because they are merely necessary conditions of the harm
su¬ered. In legal terms this is often put by saying that the court is looking for the
cause of the damage, or the ˜real™ or ˜e¬ective™ or ˜proximate™ cause, or (in legal Latin)

27 Jobling v. Associated Dairies Ltd [1982] AC 794.
Causation and remoteness of damage 119

the causa causans and not just a causa sine qua non. In other words, the court must
select one or more out of all the factors but for which the damage would not have
occurred. All of these factors are causes in fact, but this approach involves selecting
one or more of them as a ˜cause in law™. Given that there are a very large number of
necessary conditions of every event, it might seem a daunting task to have to pick out
one (or a few) of them as being ˜really™ the cause(s) of the event. However, in prac-
tice, the job is made very much more straightforward because in a tort action the
question is not, ˜what really caused the victim™s injury?™ but rather, ˜did the allegedly
tortious conduct really cause the victim™s injury?™. In fact, it is for this reason that in
the majority of tort cases, once it has been established that the defendant committed
a tort and that the claimant™s injury would not have occurred but for the tort, the tor-
tious conduct will be held also to be the (legal) cause of that injury. It is only in rela-
tively unusual cases that it will be possible to point to some other factual cause as
being a stronger candidate than the tort for the title ˜legal cause™. In most cases, all the
factual causes other than the tortious conduct will simply be the background against
which, or the surroundings in which, that conduct occurs and has its e¬ects.
What is actually meant by saying that a person™s conduct was the cause of this or
that event? The ¬rst thing to note is that this type of causal inquiry may be made
for at least two di¬erent reasons: the inquirer may be seeking an explanation of
what happened and of how it happened; or they may be asking a question about
who ought to be held responsible for what happened. Explanatory inquiries are
usually directed either at ¬nding out what has to be done in order to achieve a par-
ticular desirable result,28 or at ¬nding out how to prevent particular undesirable
things happening in the future. Di¬erent people may have di¬erent reasons for
seeking causal explanations for one and the same event, and this might lead them
to pick out di¬erent factual causes of an event as being the causal explanation for
it.29 Take a road accident, for example. The highway authority which is responsible
for seeing that road surfaces are not excessively skid-prone may be more interested
in one factor; the motor manufacturer who wants to make cars that do not over-
turn too easily will be more interested in another factor; the driver who wants to
learn how to drive without overturning the car, in yet another, and so on.
The legal causation issue in tort law is not concerned with explaining what hap-
pened or with preventing injuries in the future30 but with an ˜attributive™ question:
should we attribute responsibility for this consequence to that cause?31 If A injures B

28 Causes as recipes: T. Honor©, ˜Necessary and Su¬cient Conditions in Tort Law™ in D.G. Owen ed.,
Philosophical Foundations of Tort Law (Oxford, 1995), 375.
29 See R.G. Collingwood, ˜On the So-Called Idea of Causation™ (1937“8) xxviii Proceedings of the
Aristotelian Society 85, 92“3 for a famous passage making this point.
30 Not everyone would agree with this. Some people say that the whole point of tort law is to reduce
accidents and injuries in the future. For such people, the causal issue in tort law would involve a
form of explanatory inquiry because in their view, tort liability ought to be imposed on the person
best placed to reduce accidents and injuries in the future.
31 For this reason, it has been argued that the word ˜cause™ should be used only to refer to factual
causes, and that issues of legal causation should be thought of in terms of responsibility or scope
120 Chapter 5

and B is killed while in an ambulance being driven to hospital, the legal causation
question is: should A be held responsible for B™s death? At ¬rst sight, explanatory
causal inquiries may appear to be very di¬erent from attributive causal inquiries.
When we seek explanations, we want to know what happened, not who was respon-
sible for it or who ought to do something about it. When we think about the matter
more deeply, however, the two types of inquiry look rather similar. This is because
they are both concerned with picking out one or more necessary conditions of an
event as being in some sense ˜more important™ than the mass of such conditions
which make up the background of the causal picture. At one level, all necessary con-
ditions are equally important, exactly because they are all, by de¬nition, necessary.
For example, where a driver knocks down a pedestrian, the presence of the pedes-
trian at the site of the accident is just as necessary a condition of the incident as
the negligence of the driver. A necessary condition will only assume the foreground
if we have some particular purpose in asking the causal question that points to that
condition as being more important (for our purpose) than all the other necessary
conditions of the incident we are interested in. This is as true of explanatory inquiries
as of attributive inquiries. The di¬erence between them is that the purposes that
motivate them are di¬erent. Whereas the purpose of a highway authority in investi-
gating a road accident in which a pedestrian is injured by a driver may be to discover
whether and how improvements in road design (such as the installation of tra¬c
lights) could help to prevent such an accident in the future, the purpose of a court in
a tort action arising out of the incident will be to decide whether the driver ought to
be required to compensate the pedestrian for injuries su¬ered in the accident.
Moreover, the law™s interest in causation is crucially a¬ected by the fact, noted
above, that the question in a tort action is not ˜who should be held responsible to
pay compensation?™ but rather, ˜should the defendant be held responsible to pay
compensation?™. This question will not be answered ˜no™ simply because there is
some other factual cause which the court thinks ought to share responsibility with
the defendant, but only if the court can identify some other factual cause which, in
its view, is so ˜potent™ that it ought to relieve the defendant of any responsibility. The
car accident in which Princess Diana died provides a good illustration: it was
alleged that the driver of the car was negligent in the legal sense. It was also sug-
gested that occupants of the car might not have died if there had been crash barri-
ers in the underpass in which it occurred. In such a case, if the driver were sued, he
could not escape liability simply by pointing to the defect in the design of the road
even if, had the road authority been sued, it could have been held liable. More than
one person may be legally responsible for an accident, and it is up to the victim to
decide which person to sue. Each of the legally responsible parties may be held fully
liable to the victim, and they must argue amongst themselves about how responsi-
bility ought to be shared between them (4.6).

of liability for the consequences of factual causes: J. Stapleton, ˜Cause in Fact and the Scope of
Liability for Consequences™ (2003) 119 LQR 388.
Causation and remoteness of damage 121

Explanatory causal inquiries, then, have di¬erent purposes from attributive
causal inquiries. The criteria we use in deciding whether one thing caused another
are directly related to our purpose in making the causal inquiry. What are the crite-
ria that tort law uses to identify the legal cause(s) of injuries for the purpose
of deciding whether to impose liability to compensate for those injuries?32 There are
several which can be identi¬ed. One is that human conduct tends to be identi¬ed as
the cause of an event in preference to non-human (˜natural™) occurrences. So far as
the ˜natural world™ is concerned, the basic approach of tort law is that tortfeasors
must take the world as found (i.e. without attempting to o¬„oad responsibility onto
nature). A natural causal factor will be treated as the cause of an event in preference
to tortious human conduct only if it was su¬ciently out of the ordinary and
improbable that it could be described as ˜totally unexpected™ or a ˜sheer coincidence™.
What does the law say when conduct of someone other than the defendant is
amongst the factual causes of the victim™s injury? Such conduct may be that of the
victim or of some ˜third party™. Once again, the law™s approach is that only conduct
which is out of the ordinary will relieve the defendant of liability. Whereas a ˜natural
event™ will be treated as extraordinary if it was highly improbable, human conduct
will be treated as out of the ordinary if it was very unreasonable. For instance, in
one case a person was injured by another™s negligence and as a result he lost control
of his left leg. He fell while descending, unassisted, some steep stairs without a
handrail, and broke an ankle. The House of Lords held that the injured persons™
conduct, although ˜not at all unlikely™, was ˜quite unreasonable™; and that the tortfea-
sor was not liable for the broken ankle.33 Again, suppose that a person is injured in a
car accident as a result of a driver™s negligence, but then receives further injuries as
a result of negligence of a doctor providing treatment for the initial injuries. In such
a case, the driver will usually be held responsible for the further injuries;34 but not,

32 By far the most thorough examination of these criteria is that of H.L.A. Hart and T. Honor©,
Causation in the Law (2nd edn, Oxford, 1985), esp. chs. VI“X. For a very good summary of the
authors™ views see J. Stapleton, ˜Law, Causation and Common Sense™ (1988) 8 Oxford J. Legal Studies
111. One of the main aims of Causation in the Law was to counter the view (which was, at one stage,
widely held by scholars in the USA, but never really took hold in the UK) that the only concept of
causation used in tort law was factual causation. In my view, the authors entirely succeeded in doing
so. (Of course, whether factual causation should be the only concept of causation in the law is a
di¬erent issue: see n. 31 above.) The discussion of Hart and Honor©™s work in the ¬rst ¬ve editions
of this book focused on the authors™ assertion that legal notions of causation were derived from
˜commonsense usage™ of causal language outside the law. The main issue we addressed was whether
their account paid su¬cient attention to the distinction (which they recognized) between explana-
tory and attributive causal inquiries and to the role of purpose in choosing amongst factual causes.
Our analysis tended to assume (wrongly, as I now believe) that, unlike legal causal inquiries, causal
inquiries outside the law are typically concerned with explanation, not attribution, of responsibil-
ity. In fact, the purposes of causal inquiries outside the law are extremely various. For this reason,
I do not agree with Honor©™s view (˜Necessary and Su¬cient Conditions in Tort Law™ in
Philosophical Foundations of Tort Law, 385) that ˜the same concept of cause is used for discovering
recipes, for explaining events, and for assigning responsibility for outcomes™. Causation is much
more context-speci¬c than this statement seems to allow. It is mainly for this reason that the dis-
cussion in this edition concentrates on explaining the criteria of causation used in tort law.
33 McKew v. Holland and Hannen and Cubitts (Scotland) Ltd [1969] 3 All ER 1621.
34 But the negligent doctor may also be held liable for them.
122 Chapter 5

according to the High Court of Australia, if the medical treatment was ˜grossly neg-
ligent™.35 In that case, the doctor alone will be responsible for the further injuries.
There are three possible outcomes in such cases: either the defendant will be held
solely responsible for the further as well as for the initial injuries; or the defendant
will be held responsible for the initial injuries but not for the further injuries; or the
initial tortfeasor will be held jointly responsible with the other person for the fur-
ther injuries as well as solely responsible for the initial injuries. In the last case, if the
other person is the claimant, their damages may be reduced on account of their con-
tributory negligence; and if the other person is a third party, the defendant may be
able to recover from them a ˜contribution™ to the damages payable to the claimant.
Returning to the criteria of legal causation, these examples indicate three more.
One is that the agent of harm is more likely to be treated as its cause than the su¬erer
of harm; another is that tortious conduct is more likely to be treated as the cause of
harm than non-tortious conduct; and a third is that more-culpable conduct is more
likely to be treated as the cause of harm than less-culpable conduct.36 Suppose a
motorist knocks down and injures a pedestrian. Although both the conduct of the
driver and the presence of the pedestrian ˜in the wrong place at the wrong time™ are
necessary conditions of the pedestrian™s injuries, the driver is more likely to be
treated as their cause if neither the driver nor the pedestrian was negligent. The
driver is even more likely to be treated as the cause if the driver was negligent but
the pedestrian was not. If both were negligent, the pedestrian™s conduct is unlikely
to be treated as the sole cause unless that conduct was extremely foolish (as in
McKew v. Holland).
It is clear that these criteria of legal causation are closely related to the idea of per-
sonal responsibility for conduct. A basic function of tort law is to allocate responsi-
bility for harm, and the concepts of causation used in the law inevitably re¬‚ect
this purpose. It is interesting to consider what the relationship is between the
responsibility-oriented concepts of causation in tort law and ideas of responsibility
adopted in everyday life. Does tort law re¬‚ect widely held views about responsibil-
ity for harm or is it, by contrast, an ethical system developed by the courts with little
or no reference to what ordinary people think? There is a long history of courts
saying that issues of legal causation should be resolved on the basis of ˜common-
sense™;37 but is this the common sense of judges or of non-lawyers? These are di¬cult
questions to answer partly because of the problem of ¬nding out what ˜ordinary

35 Mahony v. J. Kruschich (Demolitions) Pty Ltd (1985) 156 CLR 522, 530. Both of these cases are
examples of what is called ˜intervening causation™ where the conduct of the injured person or the
third party occurs after the tortious conduct of the defendant and ˜intervenes™ between the defen-
dant™s conduct and some or all of the victim™s injuries. Similar principles apply to what might be
called ˜initial causation™ which concerns the allocation of responsibility as between causal factors
operating more or less at the same time.
36 This last criterion is the basis of the view of Hart and Honor© (Causation in the Law, 42) that ˜. . .
a voluntary human action intended to bring about what in fact happens, and in the manner in
which it happens, has a special place in causal inquiries™.
37 Hart and Honor©, Causation in the Law, 9. A good example is Medlin v. State Government
Insurance Commission (1995) 182 CLR 1, 6.
Causation and remoteness of damage 123

people think™ about complex issues of responsibility. However, there is empirical
research which throws doubt on the idea that legal concepts of responsibility and
causation mirror notions which are widely held outside the law.38 This research sug-
gests that personal injury victims may often sue a person, whom they do not con-
sider responsible for their injuries, simply because they know, or have been told, that
suing that person o¬ers the best hope of obtaining compensation. Conversely,
a person considered responsible may not be sued because, for example, the victim
feels this to be an unjusti¬ably aggressive act, or judges that the bene¬ts of suing
the person would not outweigh the ˜anticipated expense, trouble, upset and un-
certainties of doing so™.39 This helps to explain why accidents in the home are so
rarely the subject of litigation. People, it seems, often make (or fail to make) tort
claims for reasons having little to do with notions of cause and responsibility, and
then justify their action by use of such ideas. Furthermore, the decision to sue and,
later, the attribution of responsibility, often seem to re¬‚ect current legal rules and
the pattern of e¬ective tort liability rather than being re¬‚ected by the law. In other
words, people often sue because they think they have a good chance of success, and
they justify this decision in terms of the language and concepts of the law.
It does not follow from these ¬ndings that legal attributions of responsibility are
not based on morality or on ideas of justice. But the research does suggest that this
morality is one worked out by the judges rather than one taken by the law from the
re¬‚ections of the ˜common person™ on the sort of situations that may give rise to
tort liability. This is, perhaps, not surprising because courts are often confronted
with very unusual factual situations and very tricky ethical issues. An interesting
example is a case in which A received serious head injuries in a car accident caused
by B™s negligence. As a result, A su¬ered a personality change and turned to crime,
including rape. A was awarded damages against B, including an amount to com-
pensate him for the e¬ects of being sentenced to life imprisonment for sexual
o¬ences, on the ground that this was a compensatable consequence of B™s negli-
gence. Two of A™s victims then successfully sued him for damages for assault, which
led A to sue B again to recover damages representing the amounts awarded to the
two victims. This last action failed on the ground that the award of damages to the
two victims was too remote a consequence of B™s negligence, and that to award A
compensation would be ˜distasteful™ and contrary to public policy.40 A judge in New
South Wales has expressed the view that the decision in the ¬rst of these cases was
wrong, essentially because the moral responsibility (and, according to the criminal
law, the legal responsibility) for A™s crimes rested on A, not B, even though A would
not have committed them but for B™s negligence in injuring him.41

38 S. Lloyd-Bostock, ˜Fault and Liability for Accidents: the Accident Victim™s Perspective™ in Harris
1984 Survey, ch. 4.
39 Ibid., 157.
40 Meah v. McCreamer (No. 1) [1985] 1 All ER 367; W v. Meah [1986] 1 All ER 935; Meah v.
McCreamer (No. 2) [1986] 1 All ER 943.
41 State Rail Authority of New South Wales v. Wiegold (1991) 25 NSWLR 500 (Samuels JA).
124 Chapter 5

Examples such as this show that although legal concepts of causation and respon-
sibility are based on non-legal ideas, they are and need to be much more detailed and
complex than their non-legal counterparts. An important reason for this arises out
of the point, made earlier, that concepts of cause and responsibility serve a variety
of di¬erent purposes. An important function of legal concepts (unlike their non-
legal counterparts) is to justify the imposition of obligations that can, ultimately, be
enforced by the coercive power of the State. Furthermore, tort law is mainly con-
cerned with obligations to pay monetary compensation for personal injury, property
damage and economic loss, whereas outside the law the payment of compensation
is rarely in issue. Rules and principles on the basis of which people can be forced
to pay over large amounts of money to another must be precise, clear and sensitive
to the facts of individual cases to a degree not required in non-legal contexts.
A good illustration of the divergence between legal and non-legal concepts is the
very use of the word ˜cause™ itself. Take, for example, the case mentioned above of a
person who is negligently injured in a car accident and then further injured by the
negligence of a hospital doctor. In everyday parlance, we would probably not say that
the negligent driver ˜caused™ the further injuries “ they were ˜caused™ by the doctor.
On the other hand, even in relation to the further injuries, the negligence of the
driver could be said to be a more important causal factor than many other of the
necessary conditions of the further injuries. The role of the driver in causing
the further injuries might be described in terms of ˜creating the situation in which
the further injuries might be su¬ered™. In this case, the law is prepared to impose an
obligation to pay compensation on the basis of a weaker causal connection than is
described by use of the word ˜cause™. There are, in fact, many instances of this phe-
nomenon in tort law. Employers can be held vicariously liable for torts of their
employees on the basis that the employer provided the opportunity for the
employee to commit the tort and cause harm to another. Similarly, under Part I of
the Consumer Protection Act 1987, distributors of products may, under certain cir-
cumstances, be held liable for personal injuries caused by a defect in a product even
though they did not ˜cause™ the injuries complained of, but, at most, created the situ-
ation in which they might occur. Liability for omission provides another example “
here it can typically be said that the tortfeasor failed to prevent harm occurring, but
not that the tortfeasor ˜caused™ the harm. A di¬erent type of case is where harm
results from one person™s reaction to the conduct of another person “ for instance,
where A is induced or persuaded by B to act in a way which causes harm to C, or
where A acts in reliance on something said by B with the result that harm is caused
to A or C. In such cases we might want to say that the harm was ˜caused™ by A, even
if we want to hold B responsible for their part in producing the harm.
In all these types of case, the law is prepared to impose liability to pay compensa-
tion for harm su¬ered on the basis of a causal connection between that harm and the
defendant™s conduct which may not be easily described by use of the word ˜cause™. And
yet in law we say that these ˜weaker™ forms of causal connection satisfy the legal
requirement of liability for harm that the harm was ˜caused™ by the tortfeasor™s
Causation and remoteness of damage 125

conduct. The conclusion to be drawn is that although there are obviously connec-
tions between notions of causation inside and outside the law, the legal concept of
causation is much more complex and detailed than its non-legal counterpart; and
that in important respects, the legal concept of causation serves purposes and per-
forms functions which its non-legal counterpart does not.
It is important, however, not to jump from this conclusion to the view that cau-
sation in the law is really ˜all a matter of policy™ and that the language of causation
is used merely as a cloak for attributions of responsibility on non-causal grounds.
There are certainly concepts in tort law that can be used to relieve people of liabil-
ity for harm which they can be said to have ˜caused™. Such concepts include the ideas
of ˜scope of the risk™ and ˜foreseeability™, which we will look at a little later in this
chapter; and they include the concept of ˜duty of care™, which we examined in
chapter 3. It is also true that the decision to impose liability on the basis of weaker
causal connections than are captured by the word ˜cause™ may be based on non-
causal considerations “ vicarious liability provides an obvious example. Further-
more, it is important always to bear in mind that the way we answer causal
questions will depend on our purpose and interest in asking them. Nevertheless,
causal concepts play an independent role in tort law and cannot simply be equated
with ˜policy™. Two examples may illustrate this point. Suppose a person negligently
starts a ¬re that burns down half a town. We might well say that the negligent
person caused all the damage done by the ¬re. On the other hand, there may be
good policy reasons not to hold that person liable for all the damage caused “ such
as, for instance, that it is much more sensible for the risk of ¬re damage in crowded
urban areas to be dealt with by property owners taking out insurance against ¬re
damage to their own property.42 Or suppose that a vehicle is left by its owner
unlocked and unattended in the street with the keys in the ignition. The car is stolen
and the thief injures someone by negligent driving.43 Causal principles might point
to the thief as being (at least primarily) responsible for the harm done, while as a
matter of policy there is a good argument for imposing liability on the car owner
who is (and indeed must be) insured against such liability, who can be said to have
created an opportunity for the harm to occur and who could have reduced the risk
of the harm occurring by the simple expedient of locking the vehicle. These exam-
ples show that although the questions of whether A caused harm and of whether A
ought to be held liable to compensate for that harm are related to one another, they
are not one and the same.

5.3.2 Damage not within the risk
One of the ways in which matters of ˜policy™ can be taken into account in imposing
liability for the consequences of tortious conduct may be formulated as follows: lia-
bility for breach of a rule only extends to consequences the risk of which that rule

42 For further discussion of this example see Hart and Honor©, Causation in the Law, 89“90.
43 See e.g. Topp v. London Country Bus (South West) Ltd [1993] 1 WLR 976; J.G. Fleming, ˜Injury
Caused by Stolen Motor Vehicles™ (1994) 110 LQR 187.
126 Chapter 5

was ˜designed™ to guard against. This principle is a well-accepted one in English law
in connection with actions for breach of statutory duty. In Gorris v. Scott44 the
defendant shipowners were required by statute to provide pens for all animals
carried on board ship. When the claimant™s animals were swept overboard in a
storm, C sued the defendants on the ground that they had failed to provide pens,
which would have prevented this disaster. It was held that they were not liable
because (although failure to provide the pens was the cause of the loss of the sheep)
the ˜purpose™ of the statute was not to protect the animals against perils of the sea
but against the spread of infection.
Application of the same notion to cases of negligence may be illustrated by
reference to the decision in Doughty v. Turner Manufacturing Co.45 in which the
defendant was a manufacturer who used vats of extremely hot liquid chemicals
in its processes. These vats were protected by asbestos covers, and an employee of
the defendant replaced one of these covers carelessly so that it fell into the vat.
Unknown to anyone, the asbestos was prone, in conjunction with the heated chem-
icals, to produce a violent explosion; and this in fact occurred and injured the
claimant. Since nobody knew of the danger presented by the interaction between
the asbestos and the chemicals, C could not argue that D (or the employee in ques-
tion) should have foreseen the possibility of an explosion; but he contended that
the employee should have foreseen the possibility of some of the hot chemicals
splashing out and injuring someone; that, accordingly, dropping the asbestos lid
into the vat was a negligent act, and that the defendant was vicariously liable for the
claimant™s injuries. The defendants were held not liable on the ground that the risk
of explosion, being unforeseeable, was not a risk which the defendant™s employee
ought to have taken precautions against when the lid was replaced on the vat.
One way of stating the ground of decision in Doughty is to say that the defendant
was not negligent in relation to the risk of explosion because that risk was unfore-
seeable, and it cannot be negligent not to take precautions against an unforeseeable
risk. In other words, on the approach currently being considered, a defendant is only
liable for consequences the risk of which he or she ought to have taken precautions
against. The risks relevant to determining whether the defendant was negligent and,
therefore liable at all, are also the risks which de¬ne the extent of the liability.
Liability only extends as far as the concept of negligence itself.
This ˜risk theory™ of the extent of negligence liability has several shortcomings.
In the ¬rst place, as we will see in 5.3.3, there can be liability in negligence for con-
sequences that were not reasonably foreseeable and against which precautions
would not be required. Conversely, there are circumstances in which even fore-
seeable consequences may not fall within the scope of liability even if failure to
take precautions against the occurrence of such consequences would be negligent.
Secondly, even though the idea of ˜harm-within-the-risk™ may be able to explain the

44 (1874) LR 9 Ex 125.
45 [1964] 1 QB 518.
Causation and remoteness of damage 127

results in certain cases, by itself it does not give any guidance as to whether a par-
ticular outcome is within a particular risk.

5.3.3 Foreseeability again
In chapter 2 we saw that negligence is failure to take reasonable precautions against
foreseeable risks of harm. It cannot be negligent to fail to take precautions against
unforeseeable risks of harm. Foreseeability is one (but not the only) component of
the legal concept of negligence. Foreseeability is also used in tort law as a criterion
to determine the extent of liability for the consequences of negligent conduct. If a
consequence is very unusual, or very ˜remote™ from the tortious conduct, it may be
said that the consequence was unforeseeable and that, therefore, the tortfeasor is not
liable for it, even if it was caused (in the factual sense: 5.2) by the tort. This course
was adopted by the Privy Council in the famous Wagon Mound (No. 1) case,46 which
overturned the equally famous Court of Appeal decision in Re Polemis.47 In Re
Polemis it was held that provided some damage was foreseeable, the defendant could
be held liable for all the damage which was a direct consequence of the negligence. In
The Wagon Mound (No. 1) the Privy Council rejected this approach and purported
to lay down a similar test for the extent of liability as had already been laid down for
the existence of liability, namely foreseeability. The justi¬cation for doing this was
so that the two tests would rest on similar notions of fault: if there is to be ˜no lia-
bility without fault™, then that maxim must apply to the extent of liability as much
as to the existence of liability. However, the conclusion does not really follow from
the premise. If a person ought to have taken certain precautions, it is not obviously
unfair to hold them liable for the consequences of their failure to do so whether or
not they were all foreseeable.
In any event, as a test of extent of liability (˜remoteness of damage™, as lawyers
call it) foreseeability only requires that the type or kind of damage su¬ered be fore-
seeable, not its exact extent or manner of occurrence. For this reason, a tortfeasor
can be held liable for consequences of negligence, which are a direct result of fore-
seeable consequences, but which were not themselves foreseeable. A burn on the lip
may lead to cancer and death, and the person responsible for the burn may be liable
for the death;48 an electric shock may stimulate a latent polio virus, and the person
responsible for the shock may also have to pay for the e¬ects of the polio;49 injuries
may lead to melancholia and suicide, and the tortfeasor liable for the injuries may
also be liable for the death.50 In this way many injuries or diseases that are in a sense
merely ˜triggered o¬™ by the original negligently caused accident are brought within
the scope of the tort system, and so may be the subject of compensation. Results of

46 [1961] AC 388.
47 [1921] 3 KB 560.
48 Smith v. Leech Brain [1962] 2 QB 405; see also Warren v. Scruttons [1962] 1 Lloyd™s Rep 497.
49 Sayers v. Perrin [1966] QdR 89.
50 Pigney v. Pointer™s Transport Services Ltd [1957] 1 WLR 1121; see also Brice v. Brown [1984] 1 All
ER 997.
128 Chapter 5

this kind can no doubt be explained by the overriding desire to compensate people
for tragic misfortunes,51 especially in cases where the defendant is insured against
liability. But the element of chance or luck is very obvious in such cases. Many
people die of cancer every year; many others commit suicide; many contract polio
or other crippling diseases. Why should the majority go unaided by the tort system
while a handful, who are able to latch their disease on to some negligently caused
injury, are so generously treated? In some cases it may seem very unfair to hold the
tortfeasor responsible, yet large sums may be awarded. For example, in one case52
the injured person su¬ered a minor graze through slipping on steps that had been
negligently covered with oil. Unfortunately he su¬ered a freak reaction to an anti-
tetanus injection, with very serious results. He was awarded damages of over
£30,000 (in 1974) “ yet the only negligence consisted of leaving some oil on a step
ladder. Cases of this kind might be seen as involving liability without fault: while
initial negligence must be proved, the claimant recovers damages for what most
people would regard as nothing but an accident or a coincidence.
Whereas a tortfeasor may, under certain circumstances, be liable for unforesee-
able consequences of negligence, there may, conversely, be no liability for foresee-
able consequences of negligence if they were (partly) the result of the (foreseeable)
conduct of a human agent following the tort: ˜unreasonable™, but foreseeable,
conduct by the injured person, perhaps;53 or criminal conduct by a third party.54
Such results arise out of the feeling that in most circumstances, one person should
not be liable for the conduct of another unless the former is under some sort of
moral duty to control or protect the latter. But the law is not consistent in this
regard because courts are quite willing, for example, to impose liability, for the
results of negligent medical treatment, on the person who caused the injuries that
necessitated the treatment.
Another type of case in which there may be no liability for foreseeable conse-
quences involves economic loss, not personal injury. Suppose that a person would
not have bought a particular property “ or any property at all “ but for a negligent
valuation report to the e¬ect that the property was worth more than its actual value.
Suppose, further, that after the purchase, there is a general fall in property values.
Is the negligent valuer liable for loss su¬ered by the purchaser as a result of the fall
in property values, or only for the di¬erence between the reported and the actual
values of the property? The relevant judicial decisions are di¬cult to understand
and interpret; but the law seems to be that even if the market fall was foreseeable,
the valuer will be liable for additional loss ¬‚owing from the fall only if the report
e¬ectively stated that there would be no such fall. Of course, if the market fall was

51 But the rules may not be so generous when the loss su¬ered is purely economic; the position in
relation to property damage is somewhat unclear.
52 Robinson v. Post O¬ce [1974] 2 All ER 737. Another extraordinary case is Versic v. Connors (1969)
90 WN (NSW) (Pt 1) 33.
53 E.g. McKew v. Holland and Hannen and Cubitts (Scotland) Ltd [1969] 3 All ER 1621.
54 E.g Lamb v. Camden LBC [1981] QB 625.
Causation and remoteness of damage 129

foreseeable, it would be negligent of the valuer not to take this into account in
valuing the property. But even if the valuer was negligent in this way, the purchaser
could not recover damages for the loss resulting from the fall in the market, but only
for the di¬erence between the reported value of the property and its actual value at
the time of the valuation, taking account of the foreseeable fall in the market. This
shows that a particular foreseeable risk might be relevant to deciding whether a
person has been negligent even though, if that risk materializes, the person would
not be held liable for resulting loss.
An analogous issue might arise in a personal injury case: suppose a person would
not have gone mountain climbing but for a doctor™s negligent advice that the
climber™s bad knee would withstand the strain. Suppose, further, that in addition
to su¬ering knee damage, the climber is badly injured in a rock fall. The doctor™s
liability would probably not extend to the additional injuries despite the foresee-
ability (or even high probability) of rock falls. Similarly, doctors are required to
warn their patients of ˜material™ risks inherent in medical procedures to be per-
formed by the doctor on the patient. But the mere fact that a risk is foreseeable will
not necessarily make it material. Suppose a doctor fails to warn a patient of a mate-
rial risk, and that the patient would not have consented to the procedure if the
warning had been given; but also that the patient su¬ers harm as a result of the
materialization, not of the risk of which warning should have been given, but of a
foreseeable but non-material risk. The doctor would not be liable for the harm.
It is clear, therefore, that it is only in a very quali¬ed sense that foreseeability is
the test of the scope of liability for the consequences of which negligence is the
factual cause. A negligent tortfeasor may be held liable for unforeseeable conse-
quences of the tort, and may escape liability for foreseeable consequences of the
tort. To the extent that the ˜risk theory™ (5.3.2) of the scope of liability rests on the
fact that ˜foreseeability™ is both an element of the legal concept of negligence and
also the basic test of liability for consequences, it fails as an explanation of the scope
of liability for negligence.


. 5
( 18)